1 Animals and the Formula of Humanity: Some Background

Despite their general unenthusiasm about Kant, many animal ethicists have adopted two pieces of Kantian terminology: ‘being an end in itself’ and its opposite, ‘being a mere means’. Clark writes: “In my morality, all creatures with feelings and wishes should be thought of as ends-in-themselves, and not merely as means” (Clark 1997, 10). The “heart of human oppression of animals”, according to Donaldson and Kymlicka, lies in “turning them into a means to human ends, rather than respecting them as ends in themselves” (Donaldson and Kymlicka 2011, 88). “Moral history will come to an end when all conscious creatures will enjoy the equal right to be treated as ends in themselves and not merely as means”, says Kriegel (2013, 246). This vocabulary, adapted from Kant’s Formula of Humanity (FH), seems particularly helpful to capture how animals are morally precious for their own sake, and how human beings often fail to respond to this preciousness.

However, it is exactly the FH which expels animals from Kant’s moral universe in the eyes of most readers.Footnote 1 The formula asks us to “so act that you use humanity, whether in your own person or in the person of any other, always at the same time as an end, never merely as a means” (G 4:429.10–12). It is quite natural to assume that in the FH, Kant tells us whom we should treat how. Roughly: Persons, insofar as they share in a certain rational capacity or set of rational capacities (‘humanity’), should be protected from harm and assisted in their endeavours for their own sake, not merely as a means to further, contingent ends. At any rate, on the popular reading of the FH, it is understood as a substantive moral principle which tells us how to treat individuals. To an advocate of the rights and liberation of human beings, the FH may appear like a very attractive principle because it seems so egalitarian. It appears to ground moral concern in capacities that, supposedly, all human beings share.Footnote 2 To advocates of non-human animals, however, the FH poses a straightforward problem: Whatever ‘humanity’ may be exactly, non-human animals most likely do not share in it.

If the popular reading of the FH is correct, Kantianism is inherently and inevitably inimical to the moral claims of animals. They are excluded from moral concern by the FH, which is a formulation of the Categorical Imperative, the supreme moral principle. So the popular reading that Kantian moral concern stems from the FH has important ramifications: It makes the exclusion of animals a core feature of Kantian ethics. Hence, Chignell calls the task of including animals in the FH “the holy grail among Kantian animal advocates” (Chignell 2021, 10).

But in this chapter, I will argue that there is an alternative to the popular reading on which the FH is a substantive moral principle that fixes the scope of moral concern. Indeed, it is hard to see how it could fix the scope of moral concern. On the reading I will suggest, this purpose is served by an entirely different part of Kant’s ethical system, the doctrine of ‘obligatory ends’. On this reading, Kantian moral philosophy is not as hostile to the claims of animals as many think. At least, any remaining hostility comes from a different, more peripheral place in the system. Accordingly, including animals in the FH is not the ‘holy grail’ we ought to be chasing.

In my discussion, I focus on the FH, despite there being several other formulations of the Categorical Imperative.Footnote 3 I single out the FH for three reasons: First, in the literature it is standard to read the FH as expressing a requirement of moral concern (as Reath points out, Reath 2013, 210), more so than the other formulas. Secondly, the FH is often used to clarify the other formulas when their relation to moral concern is at issue (Wood 1999, 139; see also Franklin 2005, 39ff.). Third, the FH is the formula usually invoked to explain Kant’s exclusion of animals (see footnote 1 above).

When we focus on the FH, the basic argument runs as follows: Kant’s account of moral concern starts from the notion of humanity as an end in itself, and humanity—whatever it is—is an exclusive feature of finite rational beings. Therefore, there is no room for moral concern for animals, who are in the relevant sense ‘non-rational’. To summarise, the argument at issue is this:

Exclusion of animals by the Formula of Humanity (FH)

  1. (1)

    The FH and its supporting argument fix the scope of moral concern,

  2. (2)

    and it draws the line precisely at finite rational beings,

  3. (3)

    but animals are non-rational,

  4. (4)

    therefore, animals must be excluded from moral concern.

Most readers take it that (1)–(3) adequately represent Kant’s view and jointly explain Kant’s exclusion of animals.Footnote 4 One response on the part of animal-friendly Kantians has been to argue pace Kant that (3) is false, usually by referring to empirical work on the cognitive capacities of animals (Rocha 2015; Balluch 2016; Thomas 2016; Judd and Rocha 2017). However, it is very hard indeed to argue that non-human animals are ‘rational’ in the sense the argument requires. For ‘rationality’ in the sense at issue requires transcendental freedom.Footnote 5 That is, animals would have to be able to determine their actions independently of the laws of nature. But first, this capacity is only possible through practical autonomy, through acting on a self-imposed law. The empirical observations referenced in the aforementioned contributions of course do not show that animals have this capacity. In fact, no observation in the natural world, which is after all structured by the laws of nature, could ever show that animals have the capacity to determine themselves independently of these laws. And even if we simply posited that animals are transcendentally free, this assumption would counterintuitively imply that animals are Kantian moral agents. After all, animals would then be capable of acting on an autonomously self-imposed law, which in Kant’s philosophy is the moral law. But if an argument for Kantian moral concern for animals implies that animals are Kantian moral agents, it proves too much. Therefore, rejecting premiss (3) is not a promising strategy.

A more popular response to the above argument is to dispute (2). This is the strategy of two particularly influential voices in the debate: Wood and Korsgaard. In Wood’s view, the FH is fundamentally about expressing reverence for the value of humanity. However, Wood argues, we can express reverence for humanity even when dealing with beings who display only fragments or prerequisites of it, just as Christians can express reverence for God in the way they treat his creation (Wood 1998, 197). Therefore, Wood’s view implies, the FH draws a fuzzy line which does not strictly have to exclude animals.Footnote 6 Korsgaard argues that Kant’s argument for the value of humanity also implies the value of animality—of the capacity to pursue objects of choice. On her account, we must presuppose both these values in order to rationally follow our ordinary pursuits (Korsgaard 2004, 104, 2012, 14, 2018, 141–145). Therefore, the true line for moral concern is not to be drawn at rational beings, but at animals. Though Wood and Korsgaard certainly pursue a worthwhile goal in trying to include animals more thoroughly in Kantian ethics, there are reasons to be sceptical about the readings they presuppose in their rejection of premiss (2). I will explain these reasons in Sect. 4.3 below.

Yet another response is to agree with all of (1)–(3), but to conclude that the FH is therefore too narrow. Some argue that ‘humanity’ should be replaced by a more inclusive term such as ‘sentience’ (Franklin 2005, 35f.) or ‘consciousness’ (Garthoff 2011, 23f.). While this is a worthwhile and interesting strategy provided we believe all assumptions of the argument, it is not the strategy I intend to pursue in this chapter. For, in contrast to the previous literature, I am going to suggest that (1) is false. That is, I reject the tacit assumption that the FH fixes the scope of moral concern. The FH is, first and foremost, a principle describing the form of a good will (how a good will wills). It is not at the same time about the material of a good will (what it wills). But clearly, the question who matters morally—whom we ought to treat how—is a question about the material of a good will. For instance, whether a good will strives to make animals happy is a question about what a good will wills. This is just not the type of issue the FH, or any formulation of the Categorical Imperative, is meant to settle. My approach to the FH is not a completely new invention—it belongs in the already established camp of ‘formal’ approaches to the FH (Reath 2013). However, so far the literature has not realised the potential of such approaches for a Kantian animal ethic.

My argument rests on a distinction between concern (see Chap. 2) and what we may call esteem: a positive evaluative attitude towards beings which responds to their capacity to embody a good will. To use a metaphor, to have concern for someone is to reach out to them—to have esteem is to take one’s hat off to them. These attitudes are distinct. The attitude of concern bears intimate connections to practical benevolence and more generally to our duties towards others. Esteem does not. This is an important distinction, since it is usually thought that Kant establishes the FH by way of an argument about the value that moral agents must necessarily ascribe to themselves. The trouble, in a nutshell, is that this self-ascribed value asks only for esteem, not for concern. And so the FH cannot provide a basis for moral concern. I follow up on this argument by showing how else Kant can account for moral concern, namely in his doctrine of obligatory ends. I then conclude with a brief statement of what the FH does, if it does not account for moral concern.

2 The Esteem-Concern Equivocation

Recall the Kantian understanding of moral concern introduced in Chap. 2. On the conception I have proposed, moral concern is a matter of the duties we have towards individuals. Since Kant denies that we have duties towards animals, he excludes them from moral concern. Let us now turn to the central question: How does Kant establish that we should have moral concern for others? A standard idea among readers of the Groundwork is that Kant develops the requirement for moral concern for others out of an account of agency, or specifically of moral agency. “The rational nature exists as an end in itself”,Footnote 7 Kant writes, and he adds: “The human being necessarily represents his own existence in this way” (G 4:429.02–04). The term ‘end in itself’ is standardly understood as a term related to moral concern—to be an end in itself is to be precious or to-be-protected in a special way (Reath 2013, 202). Hence, the idea seems natural that Kant bases moral concern on the following four-step argument:

The four-step argument for moral concern

  1. (5)

    All and only rational beings necessarily represent themselves as ends in themselves.

  2. (6)

    Therefore, all and only rational beings are ends in themselves.

  3. (7)

    Moral concern is the appropriate attitude to ends in themselves.

  4. (8)

    Therefore, all and only rational beings are due moral concern.

The key terms of this argument—‘rational being’, ‘end in itself’, ‘moral concern’—have been interpreted in various ways in the literature. For instance, while some take Kant to be talking about ‘rational beings’ specifically in the sense of beings who embody a good will (Dean 1996) or who have at least “the spark of goodness” (Hill 1980, 87; see Hill 2009, 269), others take him to be talking more broadly about beings capable of setting ends (Korsgaard 1986; Wood 1999, 118). Due to such differences, the above argument captures not just one specific reconstruction of Kant’s argument, but a whole family of reconstructions. What unites this family is that, in any case, the requirement of moral concern for rational beings is developed out of an account of agency viewed from the perspective of the agent.

What matters most for present purposes is what the term ‘end in itself’ means and how it relates to an account of moral concern in the substantive sense. In the following, I am going to argue that the four-step argument commits a fallacy of equivocation. Specifically, we mean one thing by the term ‘end in itself’ when we assert that rational beings necessarily view themselves as such, and we mean another thing by the same term when we claim that all such beings are due moral concern.

One reading which clearly drives a wedge between the premisses of the argument rests on the claim that ‘end in itself’ is a descriptive term for Kant, as Sensen holds. On his view, Kant uses the term ‘end in itself’ to descriptively refer to transcendental freedom, so that ‘rational beings are ends in themselves’ is equivalent to ‘rational beings are free’ (Sensen 2011, 127–130). But if this term is purely descriptive, then there is no immediate reason why a being’s falling under it would warrant moral concern for that being, as (3) claims. Indeed, Sensen notes: “Without freedom there would be no Categorical Imperative, but by itself this does not yet justify why one should respect others” (Sensen 2011, 107). So in order for (3) to be plausible, we would have to tacitly change the meaning of ‘end in itself’ halfway through the argument, committing an equivocation.

However, most readers take ‘end in itself’ to be an evaluative term, not a purely descriptive one (see Schönecker and Schmidt 2018). In other words, the term ‘end in itself’ denotes a kind of value. Because this is the standard reading, I will not pursue the argument based on Sensen’s interpretation any further. However, even under an evaluative reading, the term ‘end in itself’ switches its meaning in the course of the four-step argument. In a nutshell, the problem is that the value we must plausibly presuppose ourselves as having qua (moral) agents is not a kind of value that plausibly demands moral concern for its bearer. To make this point in a more orderly fashion, let me expand the table of concern-related terms introduced earlier in Sect. 2.2 into Table 4.1.

Table 4.1 Terms related to moral concern and esteem

There is now a new row of terms on the right-hand side, which lists counterparts to the terms previously introduced. While those towards whom there exist duties are called moral patients, those who have duties are called moral agents.Footnote 8 What I term ‘moral esteem’ is then defined as the appropriate attitude towards moral agents.Footnote 9

Again, we can think of formal and substantive senses of each term, depending on whether we fill in specific duties or not. In either sense, however, esteem should not be confused with moral praise, which is after all only appropriate towards good moral agents. Moral esteem in the sense that matters here is appropriate even towards the scoundrel. In the formal sense, we can conceive of it as an acknowledgement and approval of the potential for moral goodness. This would presumably amount to an attitude of recognising that a moral agent is importantly distinct from all other things in the world with regard to her moral agency.Footnote 10 In a substantive sense, we can conceive of moral esteem as the approving recognition that an agent has the capacity and responsibility to fulfil a specific duty, say, to be beneficent.

What is important to note now is that the terms in each column of Table 4.1 are interdefinable or intensionally related. The relations between the terms in each column can be stated correctly and straightforwardly in analytic claims. For instance, it is analytically true that moral patients are those towards whom there are duties. Likewise, to say that a person merits moral concern amounts to saying exactly that there exist duties towards this individual, or that she is a moral patient.

Having introduced this vocabulary, we can notice a gap which will have important implications for the four-step argument. There is a semantic gap between concern-related terms and their esteem-related counterparts: The contrasting terms in each row are not interdefinable. For instance, to say that someone merits concern is not merely a roundabout way of saying that they merit esteem. To say that someone is a moral patient is not merely a roundabout way of saying they are a moral agent. And to say that someone has a duty is not a roundabout way of saying there is a duty towards that person.

The semantic gap notwithstanding, a Kantian who endorses the four-step argument could say that the terms in each row stand in certain relations of equivalence, or even identity. For example, a Kantian can argue that moral concern is required towards exactly the same beings who also merit moral esteem, or perhaps even that the two terms refer to one and the same attitude at the end of the day. Similarly, the Kantian can say that all and only moral agents are patients, and that we have duties exactly towards those beings who themselves bear duties. At this point, we do not have to deny that any of this is possible. What matters, however, is that any such claims—be they about equivalence or identity, and be they true or false—are not analytic. They are substantive claims about moral matters. And because they are substantive claims, we can ask for substantive arguments in their favour, and these arguments can turn out to be fallacies.

Fallacies of equivocation are particularly easy to commit. What happens in such cases is that we use moral terms that can switch meanings between the left-hand and right-hand columns of the table. Take, for example, the term ‘a person’s moral value’. It is natural to take this as a term belonging in the left-hand column, as one semantically related to moral concern, patienthood, and our duties towards persons. To say that a person has moral value in this sense is to say that she is precious, to be treated with concern—think of Regan’s term ‘inherent value’, for instance (Regan 2004, 241). Let us call this property ‘moral valueC’ to mark that it is intensionally linked to the notion of moral concern.

But of course, by the phrase “this person has moral value” we might also mean something else, namely that someone is ‘a person of value’. This would most straightforwardly refer to the value of a good or decent person, and so it would be more closely related to moral praise than to what I have called esteem. However, it is also possible to understand ‘a person’s moral value’ in the sense of ‘the moral value of every person’, as denoting an equal value possessed by all beings with the potential for moral goodness, even scoundrels. By stating that persons have value, we would then be saying that persons merit the recognition of their moral potential, simply due to their capacity to determine their will in a certain way (namely, independently of natural causes). Understood in this way, the term ‘a person’s moral value’ is straightforwardly related to the notion of moral agency—let us call this ‘moral valueE’ to mark that this kind of value merits moral esteem.

Thus, the term ‘a person’s moral value’ is ambiguous; it has several distinct meanings. This ambiguity alone does not need to trouble us. However, such ambiguous moral terms make it easy to commit what we can call the Esteem-Concern Equivocation. Consider an example:

The Esteem-Concern Equivocation

  1. (9)

    All and only moral agents have moral valueE.

  2. (10)

    All and only that which has moral valueC deserves moral concern.

  3. (11)

    Therefore, all and only moral agents deserve moral concern.

This argument is fallacious. And it would not be any less fallacious if we put another ambiguous moral term in the place of ‘moral value’—say, ‘dignity’. Thus writes Herman: “If moral agency is a necessary condition of dignity, and dignity is what marks something out as a moral subject, then animals are not possible moral subjects” (Herman 2018, 178). It may be perfectly plausible to say that moral agents, and moral agents alone, have a ‘dignity’ and are ‘beyond price’, if by that we mean that moral agents have a potential for moral goodness that they should never decline to realise for the sake of non-moral goods like power, wealth, and health (see G 4:393.14). But if we claim in the next breath that this ‘dignity’ obviously demands that we protect others and promote their happiness, we have tacitly switched to a different meaning of the term, committing an equivocation. Nothing about the recognition of the potential for moral goodness obviously demands that we treat others with concern. What makes this equivocation particularly tempting is that its premisses look very plausible, and in the case of (10) even analytic.Footnote 11 But equivocations do not generally make for good arguments.

Though they fail to be good arguments, equivocations are not beyond remedy either. The inference from (9) and (10) to (11) may be perfectly innocuous if we can supplement an additional premiss which links our crucial terms. There might be a convincing way to tie valueE to valueC. We might claim, for instance, that the potential for moral goodness (associated with valueE) makes someone especially precious and to-be-treated-with-concern (in the sense of valueC). However, since ‘esteem’ and ‘concern’ are not intensionally related, this additional premiss is itself a substantive claim. Exactly this premiss would have been the crux of the matter in the first place!

Alternatively, a defender of the four-step argument could claim that the term ‘a person’s moral value’ is not ambiguous but instead multifaceted. Whereas ambiguous terms have several distinct meanings (in the way the word ‘triangle’ denotes either a geometric shape or a musical instrument), ‘moral valueE’ and ‘moral valueC’ may refer to distinct facets of one and the same moral property, namely moral value. One could argue, for instance, that it is exactly insofar as an individual merits moral esteem that she merits moral concern—or even that moral esteem is identical to concern.

But mind the gap! The different facets of such a multifaceted term must themselves be joined together by a substantive background view. There must be some reason why the different facets can be meaningfully brought together under one label, why a blanket notion like ‘moral value’ is not a chimera. But then, a multifaceted term helps the argument no more than an ambiguous term. Both require a substantive background view to make the argument convincing. By way of example, consider again the term ‘a person’s moral value’. What is it about the potential for moral goodness—which has moral valueE—that demands that we treat a moral agent with concern? Why does this value give us a reason to be beneficent towards her, say?

The question might seem puzzling. It might seem obvious that no matter what kind of value we attach to whatever capacity, any valued capacity and its bearer must be protected and the capacity’s exercise must be promoted. That is enough to explain why moral valueE leads to moral valueC. But it is not obvious by any means that all value demands this treatment. There are some evaluative attitudes and associated values which call for something else than protection and promotion. For instance, consider the value of what Kant calls the “terrifyingly sublime” (Beobachtungen 2:209.15). We may be in awe of a lion’s capacity to overpower and maul a human being. This awe is an evaluative response, and it picks up on a special evaluative property of the lion’s capacities. But of course, this should not move us to protect and promote the lion’s exercise of this capacity. The terrifyingly sublime has a kind of value, but not a kind of value which demands protection or promotion.

Even in human interactions, there are cases in which we take a positive valuing attitude to some capacities of the other which does not make protection and promotion of those capacities appropriate. Take a game of chess: Here, valuing one’s opponent calls for antagonism, not benevolence. I may approvingly acknowledge, even revere, my opponent’s capacity to make her own moves in the present game, and I may value that she makes her moves rationally with the intention to win. We could even say that I value my opponent’s instrumental-rational capacities in a categorically different, uniquely elevated way compared to, say, the way in which I value my knights or my bishops. After all, without a rational opponent, there would be no reason to value any of my chess pieces at all. But of course, valuing these chess-related rational capacities of my opponent does not make it appropriate for me to let her win. To the contrary, I display respect, reverence, esteem, and so on, by restricting her choices in the game as much as I can (literally bringing her into Zugzwang) and striving to end the game in my favour as quickly as possible. Hence, there can be valuings and values which call for actions diametrically opposed to the uninhibited exercise and the flourishing of her capacities. Of course, this is due to the inherently antagonistic nature of chess, and moral life is not like chess. But that is the point: There is some story to be told about why an individual’s ‘value’ (or any evaluative property) demands what it demands. Values can demand various attitudes and actions, be it antagonism, noninterference, beneficence, or some other response.Footnote 12

So we should not treat it as obvious that the distinctive value of moral agents demands that we treat them with concern in any particular substantive sense. Tying the unique value of moral agents to the value of what is precious and to-be-protected as part of a ‘multifaceted’ term—be it ‘moral value’, ‘dignity’, ‘respect’, or something else—merely shifts the question by one step, and this is just as uninformative as the equivocation we are trying to avoid.

Equipped with the idea of the Esteem-Concern Equivocation, let us return to the four-step argument:

The four-step argument for moral concern

  1. (5)

    All and only rational beings necessarily represent themselves as ends in themselves.

  2. (6)

    Therefore, all and only rational beings are ends in themselves.

  3. (7)

    Moral concern is the appropriate attitude to ends in themselves.

  4. (8)

    Therefore, all and only rational beings are due moral concern.

There are strong reasons to suspect that this is a token of the Esteem-Concern Equivocation, such that the type of value we ascribe to rational beings in (5) and (6) is one which demands moral esteem, but the type of value at issue in (7) demands concern. After all, (5) and (6) deal with a type of value which attaches to rational beings specifically in virtue of the way in which they necessarily represent themselves. It would be intuitive to interpret this as a value we ascribe to ourselves specifically as transcendentally free beings, a value that demands that we approvingly acknowledge the capacity of rational beings to act on a self-imposed law. But nothing in this notion of esteem for ‘ends in themselves’ suggests that we should have attitudes of practical benevolence towards others, or other attitudes characteristic of moral concern in a substantive sense.

There are two immediate objections to the hypothesis that the four-step argument commits an Esteem-Concern Equivocation. The first insists that (5) and (6) already ascribe to rational beings a type of value that demands concern. This move would require us to read (5) as the claim that rational beings necessarily represent themselves as individuals who deserve moral concern. But at best, this is true for human rational beings as a matter of natural necessity, not of all rational beings as a matter of intelligibility. For instance, we can imagine rational beings who live in paradise, never requiring any beneficence from others. They would not have to think of themselves as beings who require or deserve moral concern, at least not in a substantive sense along Kant’s lines. And even among actual human beings, there can very well be individuals who think of themselves as unworthy of moral concern—for instance, Kant’s student and correspondent Maria von Herbert, who self-depreciatingly described herself as “superfluous, unnecessary” (Langton 2007, 160; see Wood 1999, 125). Hence, this objection fails, since it rests upon an implausibly strong assumption about what all rational beings necessarily take themselves to deserve.

One could also try to avoid the equivocation the other way around, by insisting that the moral concern in (7) and (8) is merely a function of the esteem we ought to have for rational beings. The most obvious way to make this argument is to say that, since we are talking about finite rational beings with needs and wishes, we must protect their life, needs, and wishes to the point that their distinctive rational capacities can come to fruition. One problem with this argument is that it has a hard time explaining why we should have concern for specific individuals, as opposed to a general attitude of maximising the exercise of valuable rational capacities, a sort of maximising utilitarianism of transcendentally free action.

Another, much more important problem is that concern (say, in the form of beneficence) does not generally make human beings more rational or free. Nor does it generally foster the moral exercise of their rational capacities. It may be the case that being too unhappy inhibits us from being good. This consideration leads Kant to accept that there is an indirect duty to secure a minimal level of happiness for ourselves (G 4:399.03–07; MM 6:388.17–30). But above a certain threshold of happiness, no strict correlation between happiness and esteem-related value exists. So why should we have benevolent or caring attitudes towards others if they are already above the threshold? Or why, to consider the opposite case, should we bother being practically benevolent towards those so far below the threshold that our beneficence cannot suffice to make them free? It would seem like we only have duties towards humans momentarily caught in a Goldilocks zone of debilitating, but not too debilitating unhappiness. This is absurd. Hence, we cannot plausibly insist that Kant’s account of moral concern is merely a matter of protecting the conditions for the exercise of esteemed rational capacities.

The upshot of the foregoing considerations is that, even if we assume that ‘end in itself’ is an evaluative term and that the appropriate attitude towards rational beings is one of valuing their distinctive capacities, we must not take it as obvious that the distinctive value of rational beings is one which calls for moral concern. If we are not careful, we move illicitly from the assumption that rational beings merit moral esteem to the conclusion that they merit moral concern. To be sure, all is not lost. But if we are to move from the conditions of agency to a requirement of moral concern, we need a bridging story.

Two classic readings of the FH appear to provide such bridging stories, namely Wood’s (Wood 2007) and Korsgaard’s (Korsgaard 1986). What makes them particularly salient is that they offer detailed accounts of Kant’s journey from the FH to moral concern. In a nutshell, Korsgaard develops an account of moral concern out of the view that we confer goodness upon our ends by choosing them, and Wood develops moral concern out of the notion of acting for the sake of humanity. I will explain why neither succeeds as a remedy for the Esteem-Concern Equivocation. In fact, they inadvertently repeat the same equivocation one step deeper into the argument. If we do not want to ascribe a fallacy to Kant, we should cease to understand end-in-itselfhood as the basis of moral concern, and we should instead pay attention to Kant’s own way of accounting for moral concern elsewhere in his system.

3 Wood and Korsgaard Against the Esteem-Concern Equivocation

To repeat, to have moral concern ‘Kantian-style’ is to have an attitude relating to our duties towards individuals. Wood takes the FH to be central in the derivation of our duties, towards both self and others (Wood 1999, 135, 140f.). This derivation of specific duties works by means of a two-premiss argument (Wood 1999, 152f.). To use Wood’s example of the duty not to give false promises (Wood 1999, 153):

Derivation of duties from the FH in Wood’s view

  1. (12)

    Always respect humanity, in one’s own person as well as that of another, as an end in itself.

  2. (13)

    A false promise, because its end cannot be shared by the person to whom the promise is made, frustrates or circumvents that person’s rational agency, and thereby shows disrespect for it.

  3. (14)

    Therefore, do not make false promises.

According to Wood, then, the FH accounts for moral concern. The FH provides all the normative force of the duty, while the “intermediate premiss” (ibid.) merely provides the subsumption of some specific action under the Categorical Imperative. Evidently, this gives the FH “radical import” (Wood 1999, 135).

Central to Wood’s reading of the FH is the contention that it is a formula about value: “Though FH takes the form of a rule or commandment, what it basically asserts is the existence of a substantive value to be respected” (Wood 1999, 141). Wood’s driving idea is that when Kant asserts that the human being “necessarily represents his own existence” as an end in itself (G 4:429.03–04), he means to say that in acting rationally, we already acknowledge the value of humanity (Wood 1999, 126). As he puts it elsewhere: “When it subjects my actions to rational guidance by an end, humanity […] involves an active sense of my identity and an esteem for myself” (Wood 1999, 119). In setting an end for myself, the line of reasoning goes, I must presuppose that I am a being capable of picking out valuable ends. What we presuppose, as a referent of our valuing attitude of self-esteem, is the value of humanity as an end in itself.

It must be said that Wood’s ‘esteem’ for humanity is not strictly the ‘esteem’ from our table in the previous section. According to Wood’s conception, we have esteem for ourselves specifically as agents, not specifically as moral agents. Therefore, this notion of esteem is not so closely related to the notion of duty, but rather to the notion of end-directed action. What matters, however, is that there is still a semantic rift between esteem and concern, and this is all the more the case for Wood’s notion of esteem.

Wood does not claim that esteem equals concern. However, he does believe that the esteem we have for ourselves gives us reasons to treat ourselves and others with concern. In Wood’s terms, the value of humanity gives us “expressive reasons for action” (Wood 1999, 141). That is, the value of humanity demands that we act in a way that expresses the right evaluative attitude. As is well known, Kant emphasises the difference between an ‘end to be produced’ and a ‘self-standing end’, and he asserts that the end in itself is self-standing (G 4:437.27). Wood interprets this as the claim that the FH asks us to “express proper respect or reverence for the worth of humanity” (Wood 1999, 141). According to Wood’s reading, then, rational beings are due moral concern because actions in line with moral concern express the right attitude towards the value of humanity.

Though Wood uses the term ‘end in itself’ uniformly, his reconstruction of Kant’s account of moral concern still commits an Esteem-Concern Equivocation. The equivocation pertains to the notion of ‘value’ at issue. We may grant that acting on our ends already expresses a certain valuing attitude for our own humanity, since we presuppose that as rational beings, we are capable of identifying good ends. But it is unclear why the value we presuppose in ourselves should demand anything over and above that we act on our ends. Why do we not express reverence for others’ capacity to choose good ends by leaving them to their own devices? After all, they will choose good ends even without our care or beneficence. So when we take it that reverence for humanity demands moral concern in a substantive sense, we have tacitly switched what we mean by ‘value’. We started with a notion tethered to a kind of esteem, and we have finished with a notion tethered to concern. Thus Wood’s Kant appears to commit an Esteem-Concern Equivocation, the ambiguous term being ‘value’.

In Korsgaard’s view, human beings qua rational beings are ends in themselves in the sense of being originators of valuable or ‘good’ ends:

When Kant says: “rational nature exists as an end in itself. Man necessarily thinks of his own existence in this way; thus far it is a subjective principle of human actions” (429/47), I read him as claiming that in our private rational choices and in general in our actions we view ourselves as having a value-conferring status in virtue of our rational nature. (Korsgaard 1986, 196)

Korsgaard’s central idea is that in setting an end, rational beings must suppose that it is good in the sense of being worth pursuing (Korsgaard 1986, 194). Indeed, our ends are good, but what makes them good is the very fact that they have been set by a “rational choice” (Korsgaard 1986, 196). Hence according to this view, rational beings are ends in themselves in the sense of being originators of the goodness of their own ends. This is why Korsgaard’s reading is often considered to be ‘constructivist’—goodness is here considered to be a feature brought into the world by the choices of rational beings. This contrasts with Wood’s ‘realist’ reading, on which rational beings merely presuppose the existence of a substantive value, but do not themselves bring it into the world.

Now for the crucial step in Korsgaard’s reconstruction: What is ‘good’ provides reasons for every rational being:

the full realization and acknowledgement of the fact that another is an end in itself involves viewing the end upon which this person confers value as good—and when one acknowledges that something is good, one acknowledges it to be “in the judgment of every reasonable man, an object of the faculty of desire.” To treat another as an end in itself is to treat his or her ends as objectively good, as you do your own. (Korsgaard 1986, 200)

Hence the fact that rational beings are ends in themselves has consequences for how we ought to treat them: We are to treat their ends as providing us with reasons for action. This seems like a mode of moral concern. Hence, Korsgaard’s reading provides an explanation why ‘being an end in itself’ relates both to the distinctive capacities of rational beings and to moral concern.

However, now it appears that we have equivocated the notion of a ‘good end’. In Korsgaard’s view, the term ‘goodness’, as it is used here, refers to an end’s feature of providing sufficient reasons for action (Korsgaard 1986, 194), but also of being “the object of every rational will” (ibid.). We can understand this as the claim that the ends set by rational beings provide ‘public’ or ‘agent-neutral’ reasons for action (see Korsgaard 2009, 191). That is, it is not merely the case that the end provides one particular rational being with reasons for action, but it is true in general that this end provides reasons for action to any rational being who has set it. An end’s ‘goodness’, according to this notion, is its reasons-giving relation to any rational being who has set the end. But of course, no account of moral concern can get off the ground with this notion of goodness alone. As Godfrey-Smith points out: “You can have a respect for my good sense without being motivated to help me” (Godfrey-Smith 2021). What we need is not merely an end’s feature of providing reasons to any rational being who has set it, but to any rational being, period. Your end should provide me with reasons, no matter whether I have adopted it. But then, the reasons provided by your ends must be ‘public’ in a much stronger sense.

One could defend Korsgaard by saying that she meant all along that the reasons generated by the ends of rational beings are ‘public’ in the stronger sense, namely in the sense that they exert normative force even over rational beings who have not adopted the end. After all, she puts great emphasis on the claim that good ends are “the object of every rational will” (Korsgaard 1986, 194). However, this turns Korsgaard’s point into an argument we already dismissed earlier, namely that a rational being can only act under the presumption that she deserves moral concern in the substantive sense. This is patently not the case, as examples of self-depreciating people such as Maria von Herbert show (see Sect. 4.2 above).

Therefore, once again, we are left with two notions, one of which clusters with esteem and the other with concern. We may grant Korsgaard the claim that a rational being can only act on the presumption that her end provides reasons for any rational being who has set it, so that in a certain sense ‘public’ reasons must be in play. But when we go on to say that the goodness of our ends accounts for the requirement of substantive moral concern, we have taken a leap. We are now talking about ends that are ‘good’ in the sense of providing reasons even to beings who have not set them. Merely insisting that both are aspects of ‘goodness’ will not do—mind the gap. Otherwise we are merely reiterating the Esteem-Concern Equivocation.

To conclude, interesting and illuminating as they are, neither Wood’s nor Korsgaard’s reading of the FH gives a compelling account of Kant’s journey from agency to moral concern. In the attempt to employ these readings of the FH to provide a bridging story, we have merely run into the Esteem-Concern Equivocation put in different terms. So the arguments which these readings ascribe to Kant end up being fallacious. Although one could always press philosophical arguments further, I take it that these difficulties are serious enough for us to consider an alternative reading of how Kant arrives at his conception of our duties to others.

4 Obligatory Ends: How Kant Derives Duties to Others

The way forward is to dispense with the popular idea that the FH encapsulates, or is very intimately related to, Kant’s account of moral concern. One reason for this is that, as I have just argued, if we want to read an argument for moral concern into the passage leading up to the FH, that argument will not be very compelling. Another reason is that treating individuals a certain way, and so treating individuals with moral concern, inevitably has to do with effecting change in the world. To tell a moral agent how to treat individuals is to tell her what things to create, what processes to initiate, what states of affairs to bring about, in relation to those individuals. In Kant’s terms, we are dealing with the prescription of ends to be produced (G 4:437.25). But as Kant makes abundantly clear, humanity as an end in itself is not an end to be produced, but a self-standing end (G 4:437.27). This should make us suspicious of any suggestions to the effect that Kant’s account of moral concern derives from the notion of end-in-itselfhood.

However, there is yet another reason to disbelieve that Kant’s account of moral concern is to be found in and around the FH: Kant provides a better account of moral concern elsewhere, in his doctrine of obligatory ends. The goal of the present section is to sketch Kant’s argument for the requirement of moral concern. The last section will then be devoted to the question what the FH is there for, if not to provide a basis for moral concern.

When Kant introduces the notion of an obligatory end, he makes it clear that he is talking about ends to be produced: “An end is an object of free choice, the representation of which determines it to an action (by which the object is brought about)” (MM 6:384.33–34, emphasis removed and added). Kant then goes on to present arguments for two crucial claims: First, in general there must be an end that is also a duty. This is an end to be produced which we ought to adopt. Secondly, there are two obligatory ends in particular, namely one’s own moral perfection and the happiness of others. These arguments jointly provide the basis for Kant’s account of moral concern.

Kant’s first argument is presented in a brief section of the introduction to the Doctrine of Virtue. Apart from the characterisation of ends as objects of choice whose representation determines action, Kant begins by explaining the notion of an obligatory end in more detail. He emphasises that “having any end of action whatsoever is an act of freedom” through which a rational being determines herself (MM 6:385.03–04). What is important here is that nature does not set all ends for rational beings (though it does set the end of our own happiness, see Sect. 2.3 above). Considered purely as rational beings, we are capable of setting ends for ourselves freely. The laws that govern the free setting of ends are a rational being’s practical principles, not the laws of nature. Now the idea is that some principles command ends categorically, not merely as means to further ends. And an obligatory end, or “end that is also a duty” (MM 6:382.06–07), is an end that is commanded categorically in this way.

Having explained the notion of an obligatory end, Kant goes on to argue that there is indeed such an end: “Since there are free actions there must also be ends to which, as their objects, these actions are directed” (MM 6:385.11–12). Taken by itself, this only establishes that every free action has some end, not that there is one and the same particular end that is prescribed for every free action.Footnote 13 Kant however insists that “among these ends there must be some that are also (i.e., by their concept) duties” (MM 6:385.12–14). The reason for this is that autonomous action could not get off the ground without an obligatory end: “For were there no such ends, then all ends would hold for practical reason only as means to other ends; and since there can be no action without an end, a categorical imperative would be impossible” (MM 6:385.14–17).Footnote 14 The argument, as I see it, consists of two major parts: The first part is a view about free actions, namely that they exist, and that they are directed towards some end which is provided by practical reason, not by nature. The second part is a view about practical reason, namely that it determines action by means of principles which can either be hypothetical or categorical. Hypothetical imperatives have the form ‘if X is your end, do Y’ (see G 4:441.10–11), and hence practical reason can only use these principles to initiate action if some end (namely X) is already presupposed. Furthermore, since ends are often in turn means to further ends, hypothetical imperatives form chains or strings. These strings have to end somewhere if pure reason is to be practical; otherwise practical reason would merely be in the business of selecting means to presupposed ends. Now Kant’s view, already established in the Groundwork, is that there is also a Categorical Imperative, which demands actions or maxims without reference to any presupposed ends. Only thanks to this imperative is it possible for pure reason to initiate action at all. Hence, putting the two parts of the argument together, we get the conclusion that pure reason must supply at least one end of its own; otherwise it could not be practical. And if an end is provided by pure practical reason, that is an obligatory end. Therefore, there must be an obligatory end.

Kant concludes his section by emphasising again that the obligatory end is provided by reason alone, not nature. What matters for Kant’s account of moral concern, however, is that he has set the stage for materially specific moral injunctions about what rational beings ought to do. That includes what they ought to do to themselves and others. Hence, the stage is set for moral concern.

The second argument Kant offers is that one’s own moral perfection and the happiness of others are the obligatory ends. This argument is admittedly somewhat obscure, and the presentation is lamentably brief. On the face of it, Kant simply posits that one’s own perfection and the happiness of others are obligatory ends without any preceding argument (MM 6:385.31–32). He does not seem to expect the reader to have any objections or questions regarding this point. However, his remarks in the following three paragraphs in the introduction to the Doctrine of Virtue do provide some basis for his claim.

Kant’s strategy is not to argue positively for the claim that these two ends are obligatory, but instead to argue merely that the two converse ends—one’s own happiness and the perfection of others—cannot possibly be obligatory ends. One’s own happiness does not qualify as an obligatory end because it is “an end that all human beings have (by virtue of the impulse of their nature)” (MM 6:386.01–02; see Sect. 2.5), and “what everyone unavoidably already wants by himself, does not belong under the concept of duty” (MM 6:386.03–05). In other words, an obligatory end cannot perfectly map onto what we want as natural beings, or else it would lack its character of prescription. The problem with the perfection of others, by contrast, is that “it is self-contradictory to require (make it my duty) that I ought to do something that no other than he himself can do” (MM 6:386.12–14). Kant’s point is that perfection comes from setting moral ends and acting from duty, both of which we can only do for ourselves. It is therefore strictly impossible for us to promote the perfection of others. Though there are many small questions to ask about these arguments, we should focus on the big methodological issue: How is Kant’s process of elimination supposed to work by which he determines the obligatory ends? How, in particular, did Kant arrive at the initial two candidates, perfection and happiness?

By and large, the consensus among Kant readers is that Kant derives the two candidates from his account of the highest good in the world (as Allison points out, Allison 1993, 18). The highest good in the world, Kant lays out in the second Critique, consists in complete virtue together with proportionate (complete) happiness (CPrR 5:110.31–111.05). Though it is not so clear in the Doctrine of Virtue that Kant is relying on this doctrine to derive the obligatory ends, he makes it quite explicit in the Gemeinspruch, published four years earlier:

the need for a final end assigned by pure reason and comprehending the whole of all ends under one principle (a world as the highest good and possible through our cooperation) is a need of an unselfish will extending itself beyond observance of the moral law to production of an object (the highest good). (Gemeinspruch 8:279FN)

Kant’s point here is directed against his fellow philosopher Garve, who had objected to the Groundwork that a will determined by a formal principle alone (in particular, a will that determines itself independently of all considerations of happiness) could never have an object, thus it would be indeterminate what a good will wills (Garve 1802; Gemeinspruch 8:278f.). In response, Kant explains that the good will ‘extends’ itself (Gemeinspruch 8:278FN) to include an intended product (though that product is not what gives the good will its moral worth). And Kant makes it clear that the way to find the intended product of a good will is to take the highest good in the world and leave out the bits that are unattainable through our actions (Gemeinspruch 8:279.19–22). This matches with Kant’s process of elimination in the Doctrine of Virtue. Thus, it is reasonable to assume that Kant really does rely on his doctrine of the highest good in the world when arguing for the two initial candidates for obligatory ends.

At first sight, Kant’s reliance on the doctrine of the highest good in the world poses a problem. On the one hand, the doctrine of the highest good is in itself a fairly controversial part of Kant’s system. Allison remarks that it is “somewhat disappointing” to learn that Kant relies on the doctrine of the highest good to establish the obligatory ends (Allison 1993, 18). The greater difficulty for present purposes is however that Kant’s conception of the highest good, both in the person and in the world, relies on the idea of “happiness distributed in exact proportion to morality” (CPrR 5:110.33). This would suggest that Kant must exclude animals, who after all never morally deserve their happiness in virtue of embodying a good will.

To my mind, there are two acceptable ways of resolving this difficulty. The first and stronger way has already been suggested in the literature: As Cholbi points out, Kant never claims that happiness is not a good, but rather that it is a good conditioned by morality (Cholbi 2014, 348). But happiness as a good is only so conditioned because it can corrupt the wills of human beings (ibid.). This is why Kant asserts that an impartial spectator could take no delight in seeing wicked people happy (G 4:393.19–22). But animals are crucially unlike wicked people. They cannot be morally corrupted by their happiness, since they are not moral agents in the first place. Though the happiness of animals is of course never deserved through virtue, it is also never undeserved through vice. In this way, their happiness is a more innocuous good than the happiness of human beings. If the highest good in the world is the greatest possible presence of goods (conditioned and unconditioned), then we should consider the happiness of animals to form an important part of it.

Note that this reliance on the doctrine of the highest good does not represent an endorsement of teleology (see Trampota 2013, 141). Good actions are still good in virtue of springing from a good will, not in virtue of contributing to the highest good in the world. Kant here uses the notion of the highest good as a heuristic or filter, as it were, to determine what a good will would want. It would want what any rational being could endorse as a good, which is the highest good in the world. But Cholbi’s crucial point is that any rational being could very well endorse as a good the happiness of beings that are morally incorruptible.

The second way to resolve the difficulty is to circumvent the doctrine of the highest good in the derivation of the obligatory ends altogether. We can do this if we focus on the practical role of happiness and perfection as ends of human beings. In Kant’s view, happiness and moral perfection are the only two ends human beings can pursue at the end of the day. In general, setting ends is a task of practical reason according to Kant (MM 6:385.01–04). It can accomplish this task in either of two ways: first, as pure practical reason, by supplying its own end; secondly, as instrumental practical reason, by choosing the means to the end nature has imposed on human beings, their own happiness. Practical reason implements these ends and means in action by means of practical principles, hypothetical or categorical. To repeat, practical principles interact in certain ways—hypothetical principles string together, so that each end is a means to a further end. Ultimately, human beings only ever act either for the sake of the end imposed upon them by nature, or for the sake of the end they set for themselves purely as rational beings. Insofar as human beings act from the ends imposed by nature, they aim for happiness. And insofar as they act from an end they set purely as rational beings, they aim for moral perfection. This argument then relies on Kant’s background assumptions about human agency, but not about the highest good.

We can roughly summarise the proposed line of reasoning in the following argument:

Argument for the two obligatory ends

  1. (1)

    There must be at least one obligatory end.

  2. (2)

    Any end that a human being sets is in turn a means to either happiness or perfection.Footnote 15

  3. (3)

    Therefore, happiness and perfection are the two only ultimate ends (ends that are not means to any further ends).

  4. (4)

    Obligatory ends are ultimate ends.

  5. (5)

    Therefore, happiness and/or perfection are the only candidates for obligatory ends.

  6. (6)

    One’s own happiness and the perfection of others cannot be obligatory ends.

  7. (7)

    Therefore, one’s own perfection and the happiness of others are the only candidates for obligatory ends.

An obvious problem with this argument is that it cannot establish that our own perfection and the happiness of others are obligatory ends. It can only establish that they are the only remaining candidates. What is more, there is no guarantee that no further distinctions must be drawn, so that only certain instances of our own perfection are obligatory, and likewise only certain instances of the happiness of others. Indeed, Kant himself does introduce further specifications. He holds that it is only the happiness of others in the sense of their permissible ends that we ought to promote, not in the sense of their impermissible ends (MM 6:388.05–08).Footnote 16 Thus, the two obligatory ends do admit of further specification. What we can conclude from this reading is that the status of our own perfection and the happiness of others as obligatory ends has the character of a default status or a defeasible presumption.Footnote 17 The question we should ask now is not whether these ends are obligatory, but in which respects they cannot be. Hence what Kant achieves by the argument is not that he strictly proves these ends to be obligatory, but that he shifts the burden of argument.

How do we get from these two obligatory ends to moral concern? Moral concern ‘Kantian-style’, I have suggested, is an attitude of recognition of, and resolve to do, one’s duties (or some specific duties) towards individuals (see Sect. 2.1). Kant’s argument up to this point aimed to show that there are two obligatory ends, namely one’s own perfection and the happiness of others. What still needs to be explained is how the obligatory ends lead to our duties towards self and others. To reiterate a point from Chap. 2, moral concern comes in formal and substantive variants. It can be recognition and resolve with regard to one’s duties towards someone in general, or recognition and resolve with regard to some specific duties towards that individual. So the two questions we should ask at this point are thus: How do obligatory ends lead to duties ‘towards’ someone? And how are we to derive the specific duties towards self and others which Kant thinks we have?

The first question—how we arrive at duties ‘towards’—once again leads us into territory without consensus. But regarding this issue, there is also little debate. Here is my preferred reading. Kant does not explicitly offer a route that leads from obligatory ends to duties-towards. He appears to assume that we are already there. To repeat, obligatory ends are duties (MM 6:382.06–07). And Kant appears to conceive of them as duties towards self and others. That is, the promotion of one’s own moral perfection is a duty to self, and the promotion of the happiness of others is a duty to others.

But assuming the obligatory ends are already duties ‘towards’ individuals, what makes them so? There are two basic explanations, both of which have a basis in Kant’s writings. The first is that a duty ‘towards others’ is just a duty derived from the obligatory end of the happiness of others, and a duty ‘towards self’ is just a duty derived from the obligatory end of one’s own self-perfection. Thus, the material of duty (the end we should adopt) determines whether it is a duty towards others or self. This fits well with the fact that according to Kant, our duties to self and our duties to others generally differ in what they demand of us. The second way of explaining the directionality of duties concerns the form or ‘ground’ of duty. Kant clearly adheres to the view that there can only be duties ‘towards’ those whose will, in a certain sense, “this constraint” us (MM 6:241.15–17; see MM 6:442.10). One aspect of this constraint is that the individual towards whom a duty is directed can always “release” the duty-bearer (MM 6:417.19). Therefore, if we ask whether the directionality of duties is a matter of the duties’ material or form, Kant would answer yes to both. At any rate, the directionality of duties enters the stage together with obligatory ends, simply because these ends are already duties ‘towards’ someone. This gives us moral concern in the formal sense.Footnote 18

As a preliminary upshot, Kant’s account of moral concern for others (though not for ourselves) revolves around promoting happiness. To include animals in Kantian moral concern is to include them in this happiness-promoting project in the right way. Although we have not yet considered in detail how the inclusion of animals might be achieved, we can already conclude that the line of moral concern can at most be drawn at the capacity for happiness. Animals who are incapable of experiencing hedonic states will inevitably remain excluded.

One can legitimately worry, however, if the obligatory ends comprise all we usually expect from Kant in terms of moral concern. What is often considered Kant’s most important legacy in interpersonal ethics is the notion that others are not to be treated as mere means. How is this notion supposed to come out of the obligatory end of the happiness of others? Are we not rather left with a crude maxim of happiness maximisation? In the rest of this section, let me explore some things that follow from the doctrine of obligatory ends, and some that do not, regarding our treatment of others.

Among Kant’s readers over the years, the notions of ‘treatment as an end in itself’ and ‘treatment as a mere means’ have taken on a life of their own, admitting of several popular interpretations. If we think of treatment as an end in itself as a treatment to which the recipient could rationally agree, given all information and sufficient time for reflection, then Kant’s doctrine of obligatory ends gets us fairly close. It would certainly be in the patient’s best self-interest to agree to the promotion of their own happiness—though perhaps it would still be rational for the patient to disagree with such treatment on moral grounds, since her duties are different from ours.

However, some interpretations of ‘treatment as an end in itself’ and ‘treatment as a mere means’ are more ambitious than what the doctrine of obligatory ends can give us. Take, for instance, the idea that the good of individual ‘ends in themselves’ must never be sacrificed for the aggregate good of many. In animal ethics, this absolute opposition to moral trade-offs is probably the idea most strongly associated with the labels ‘Kantianism’ and ‘deontology’. Thus, Regan claims that “we fail to display proper respect for those who have inherent value whenever we harm them so that we may bring about the best aggregate consequences” (Regan 2004, 249). It is by no means obvious that so strong an idea follows from Kantian moral concern, understood along the lines I suggest. Moral concern as the recognition of our duties, and as the resolve to fulfil them, requires simply that we observe our duties. Certainly, our duties towards others rule out harming them for the sake of our own happiness, since our duty is to promote the happiness of others even at the expense of our own. But to what extent it rules out sacrificing others for the happiness of yet others is a completely different question. In any case, we should not expect an answer to this question to follow from Kant’s account of moral concern alone. If a prohibition of trade-offs is part of the Kantian picture at all, it must originate somewhere else in the system, particularly in his views on the resolution of (apparent) moral conflict.Footnote 19

To summarise, Kant’s account of moral concern rests on his doctrine of obligatory ends. The kind of substantive moral concern we get from Kant’s account may be in the ballpark of some popular ideas about ‘treatment as an end in itself’. Other ideas that are often associated with Kant do not follow, specifically a blanket prohibition of moral trade-offs. What is more, moral concern has only little to do with Kant’s actual notion of ‘ends in themselves’. Considered as objects of moral concern, we are ends to be produced. Things ought to be changed about us and our circumstances. In what sense we are ‘ends in ourselves’ is another matter, largely independent of our moral patienthood. At this point, one can legitimately wonder what the notion of an ‘end in itself’ is there for, if not to ground moral concern. Let me propose an answer in the following, final section.

5 What Is the Point of the Formula of Humanity, if Not Moral Concern?

So far in this chapter, I have rejected the popular idea that the FH is a solid basis for an account of moral concern. I have then addressed the worry that there is no other basis for moral concern in Kant by discussing the doctrine of obligatory ends. Another possible worry is now that my reading leaves the FH out of work. What, if not moral concern, could a principle be about, if it tells us to treat ourselves and others ‘as ends in themselves’, never as mere means? This is the topic of this section. My reading is only a variation over a theme already known from the literature. So my goal is simply to position my view in the debate and outline its main claims.

Reath (2013) has proposed a helpful ordering of different readings of the FH. On the one far end of the spectrum belong readings that see the FH purely as a substantive moral principle, usually a principle about moral concern. On the other far end stand those who read the FH purely as a ‘formal principle’, by which Reath means:

the internal constitutive principle of a domain of cognition or rational activity. It is the principle that defines or describes and makes it possible to engage in that activity, thus the principle that any subject engaged in that activity must follow. (Reath 2013, 204)

When we read any formula of the Categorical Imperative as a formal principle, we view it as stating a constitutive rule of a rational capacity. As Kant points out in his overview of the different formulas (GMS 4:436.08–437.04), each formula emphasises a structural feature of the maxims of a good will: The Formula of Universal Law highlights that they have a certain form (that of a candidate for a universal law), the FH highlights that they have a certain matter (namely, humanity), and the Formula of Autonomy highlights that all these maxims stand in certain systematic relations to each other, forming a “realm” or “kingdom” akin to the realm or kingdom of nature structured by laws of nature. Thus, each formula points a spotlight at some feature of a good will, of practical autonomy. Kant himself claims that the Categorical Imperative “as such only asserts what obligation is” (MM 6:225.06–07).Footnote 20

Now, Reath would strongly prefer a reading which brings together formal and substantive aspects of the FH (Reath 2013, 210). That is, he would rather interpret the formula as telling us both something about the nature of practical autonomy and how we ought to treat others. In his eyes, it is simply a desideratum for a plausible reading of the FH that it accommodates the “standard intuitive reading” of the FH as a substantive moral principle about concern (Reath 2013, 208). However, if we appreciate what the doctrine of obligatory ends does for Kant’s account of moral concern, we simply do not need the FH to be a substantive moral principle. If we want, we are free to understand the FH and the other formulas as purely formal principles of practical autonomy, with all the substantive action-guidance being provided by the doctrine of obligatory ends later on.

So what does the FH tell us about practical autonomy? Recall the main formulation of the FH: “So act that you use humanity, whether in your own person or in the person of any other, always at the same time as an end, never merely as a means” (G 4:429.10–12). The term ‘humanity’, I take it, refers to a rational being’s capacity to act on a self-imposed law, rather than merely according to natural laws. This is practical autonomy, and the self-imposed law is the moral law. What the FH tells us, in a nutshell, is that we must take this capacity itself into view if we are to exercise it.

This is not to say that we must simply contemplate autonomy or the moral law in order to act autonomously. Rather, we must have autonomy in view as an end—an object with regard to which we determine our action (see MM 6:381.04–06). To be sure, it is not an end like any other. Many of our ordinary ends are external things we aim to produce through our actions. For instance, I might use the coffee maker as a means, a cup of coffee being my end. Unlike the future cup of coffee, autonomy is neither external to the agent nor is it a yet non-existent thing. It is already there and inherent in the very type of end-setting at issue. It is not an ‘end to be produced’ (G 4:437.26–27). Still, autonomy is an ‘end’ in the more general sense of an object with regard to which our action is determined. Of course, any particular maxim of a good will also has an end to be produced—for example, the happiness of some particular person, or the cultivation of some of the agent’s own capacities that are serviceable to morality. But if we want to know more about these kinds of ends of a good will, we should not consult the FH, but the doctrine of obligatory ends. On the purely formal reading, the FH tells us something about one kind of end of a good will, but not everything about all kinds of ends it adopts.

Kant’s point, as I read it, is that we can only act autonomously if we take autonomy into view as an end, namely by setting out to exercise autonomy well. Autonomy is the capacity to act on a self-imposed moral law. This capacity comes with its own standard, namely the moral law itself. And so we exercise autonomy well only if we set out to follow the moral law. So to treat humanity as an end, in effect, is to strive to follow the moral law. By contrast, to use humanity as a mere means is to use the capacity to act on a self-imposed law for any other purpose than to actually follow that law.

Here is a good place to provide an example, and the obvious choice is Kant’s case from the Groundwork, the prohibition of prudential suicide. Kant imagines a man who is “someone who has suicide in mind” (G 4:429.16). In other words, this person is considering suicide. He is not acting on blind impulse but contemplating different ends which he could adopt. Some of these ends stem from nature, and others from reason. The opposition, it is quite clear for Kant, is one between inclination-based ends which demand suicide (to avoid greater future suffering) and ends autonomously set by pure reason which demand continuing an unpleasant life (in order to remain capable of fulfilling one’s duties). The man who contemplates suicide is aware of this opposition. But in order merely to contemplate the ends he could adopt from pure reason, the man has already had to impose the moral law on himself. He can cognise the moral law only by self-imposing it. And so he must do what the moral law demands, which is to refrain from suicide. Beings with the capacity to act on a self-imposed moral law must strive to exercise this capacity well, to actually follow the moral law. To contemplate the options and then do what inclinations demand is to use the capacity merely as a means for our own happiness. This is what the example illuminates.

To be sure, Kant’s discussion of the four Groundwork examples also presents some difficulties for a purely formal reading. First, Kant also speaks of human beings as ends in themselves, not just humanity (e.g. MM 6:429.03–04; MM 6:429.20–23). This is somewhat misleading. Evidently, what Kant here means by ‘human beings’ are individuals insofar as they are capable of acting on a self-imposed law. At the end of the day, it is more straightforward to say of the capacity that it is an end in itself (that we must strive to exercise it well). To say the same of the bearer of the capacity is merely derivative.

A second problem is more serious: Kant also talks about humanity in the person of others. It too should be treated as an end in itself, never as a mere means. In general, I understand this as an emphasis on the fact that humanity as a capacity is not tied to particular individuals but is the same capacity in all autonomous beings. However, it cannot be denied that Kant himself appears to read an awful lot into the notion of treating others’ humanity as an end in itself, especially in his discussion of false promises. He seamlessly begins to comment on whether others “can contain” the same end and takes this as a marker of whether humanity is being treated as an end (G 4:429f.). What is going on?

On my preferred reading, Kant is taking a detour here for the sake of adding some variety to the discussion. His overall goal in the examples is, still, to illustrate an aspect of morality using examples of obvious immorality. The situation in the false promise example is, once again, someone who contemplates different actions, one of which is giving a false promise in order to obtain some money. As in the case of prudential suicide, what should be shown is that in using autonomy to contemplate options, Kant’s agent already subjects himself to the moral law. Humanity is inherently a capacity to be exercised well. So Kant could simply repeat his point from the suicide example: Autonomy inherently demands that we follow the moral law. And the agent in the example already knows what the moral law demands, namely not to make a false promise. In other words, Kant could also have emphasised that the lying promiser is using humanity in his own person as a mere means.

Instead, however, Kant chooses to emphasise humanity in the person on the receiving end of the action. I suggest his move is the following: A will who exercises autonomy well sets an end from pure reason, in accordance with the moral law. And if an end is set from pure reason, any rational being in any natural circumstances could have set it. This even includes those who stand to be harmed by the proposed action, such as the recipient of the false promise. That this person could not have set the same end shows that the false promiser’s end cannot be set from pure reason. It must be set from inclination. And so by contemplating his options and opting for an inclination-based end, he has failed to treat humanity as an end in itself.

Although Kant’s discussion of the four examples puts a purely formal reading of the FH to the test, formal readings also come with a general advantage: As Reath points out (Reath 2013, 224), formal readings make it much easier to account for the equivalence of the different formulations of the Categorical Imperative. The problem is familiar in the Kantian literature that the various formulas appear to give divergent moral guidance if we interpret them as substantive principles (Geiger 2015). It would appear as if the Formula of Universal Law, the Formula of the Law of Nature, the FH, and the Formula of Autonomy all make divergent substantive prescriptions, so that cases abound in which some formulas are observed, but not others. Formal readings avoid this difficulty altogether, since each of the formulas contains no substantive moral guidance from which the others could diverge.

What about Kant’s argument that every rational being inevitably views her own existence as an end in itself? To cut a long story short, I want to suggest we are not truly dealing with an argument here. Kant is making a comment about the reflexivity of pure practical reason. In the passage leading up to the FH, Kant asserts about the Categorical Imperative: “The ground of this principle is: the rational nature exists as an end in itself. The human being necessarily represents his own existence in this way” (G 4:429.02–04). Now if we interpret the term ‘end in itself’ along the lines I suggest, Kant is saying that any rational being views her own autonomy as a self-standing end—something to be used well, by its own inherent standard, which is the moral law. Therefore, Kant’s comment simply states that we already regard our own autonomy as tied to the moral law.

Kant then adds that “every other rational being also represents his existence in this way consequent on just the same rational ground that also holds for me” (G 4:429.05–07). To this he attaches a footnote that has puzzled many readers: “Here I put forward this proposition as a postulate. The grounds for it will be found in the last Section” (G 4:429FN). Why does Kant think he has not supported his assertion enough? The answer is that up to this point, Kant has only asserted that the rational being, from a first-person standpoint, takes her autonomy to be tied to the moral law. The capacity to act on a self-imposed law rather than merely natural laws always presents itself to us as something to be used well. But it could present itself thus even if morality were a mere phantom or chimera (G 4:445.08). The assertion Kant then makes, namely that myself and other rational beings have a rational ground for viewing our own autonomy in this way, is stronger. This stronger assertion requires that we show that morality is not a chimera, that we can indeed act on a moral law as opposed to natural laws. We must show, in other words, that the will really is free. And that is what Kant sets out to show later, as promised, in Groundwork III.

To summarise, on my preferred reading, the FH is a purely formal principle of autonomous willing. It tells us that as soon as we exercise our capacity to act on a self-imposed law in any way, we are bound to exercise it well, thus to actually follow the moral law. As soon as we use this capacity to another end than to follow the moral law, we use it as a mere means. Thus we fail by a standard already internal to autonomous willing itself. To be sure, this section has not shown that only a purely formal reading of the FH can be coherent and fruitful—just that it too can be. Adopting a purely formal reading of the FH shifts our perspective away from the question whether animals are ‘ends in themselves’ and towards the question whether their happiness should be considered a part of the obligatory end of the happiness of others. This is the kind of reading we need in order to include animals in Kantian moral concern.

Between Kant and today’s animal advocates, there are some disagreements. What they need not quarrel about, however, is the FH. Contrary to popular opinion, I have argued that the FH cannot be the basis for moral concern that many take it to be. The problem is that the substantive value the FH would establish (if it were to establish any) would have to emerge out of the conditions of agency—out of how “[the] human being necessarily represents his own existence” (G 4:429.02–04). But even if we grant that human beings ascribe to themselves some substantive value, it cannot be a value that accounts for moral concern. The value will demand that we have esteem for the agent. But this is not the same as demanding concern. We can take our hat off to someone, and even be in awe of their capacity to act from the moral law, but still see no reason whatsoever to reach out to them with moral concern. The influential readings by Wood and Korsgaard cannot bridge this gap without equivocating esteem- and concern-related terms further down the line.

I have offered an alternative account both of how Kant arrives at moral concern and of what the FH is there for. Kant establishes moral concern with his doctrine of obligatory ends. The two obligatory ends are one’s own moral perfection and the happiness of others. All of our materially specific duties are derived from these ends. The philosophically crucial moves lie in establishing that there must be obligatory ends at all and that they are specifically the two ends Kant mentions. If we grant Kant these moves, however, moral concern is established. The FH ends up playing no special role for moral concern. Its purpose is entirely different and lies in illuminating the good will. In a nutshell, the FH tells us that autonomous willing inherently calls for the observance of the moral law. Humanity is the capacity to act on a self-imposed law, and we can only exercise this capacity if we reflectively take it into view (treat it as an end of our action) by striving to exercise it well. This requires that we set out to actually follow the moral law. We must not use our humanity merely to contemplate possible options and then go with what our inclinations favour. On this reading, the FH is a formal principle of an autonomous will, not a substantive principle about moral concern.

The main move of this chapter has been to set an issue aside. It may seem as though this is a mere preliminary to the real discussion, and in a way it is. But this preliminary is important. At the end of the day, the question at stake here is whether the exclusion of animals from moral concern follows from the very core of Kant’s ethical system. The FH is, after all, a formula of the Categorical Imperative, the supreme moral principle. The most important message of this chapter is thus: No, the exclusion of animals is not a necessary feature of the supreme Kantian moral principle. The issue lies elsewhere.